How master data management supports health information exchange
Recently, I had the pleasure of co-presenting “Trust in Regional Exchange Supports Patient-Centered Research” at Healthcare and Information Management Systems (HIMSS15) in Chicago, Illinois. Despite having the worst timeslot one can get—last day, last time—we had good attendance and good questions, so I’m reading that interaction as validation of the topic’s importance.
Tom Check, chief executive officer (CEO) of Healthix, the largest regional data exchange in New York state, and I explored how master data management (MDM) supports research, care coordination and health information exchange. Here’s a quick synopsis of points we made and questions we answered.
Why did Healthix and the New York City Clinical Delivery Research Network use probabilistic matching?
I shared studies IBM has done that show a deterministic approach to patient matching—a byte-for-byte comparison of attributes—will only find 30–40 percent of the potential matches, at best. Some organizations may build rules on top of the deterministic approach, but even this strategy only gets the matching up to approximately 50–60 percent, if the data is clean and doesn’t represent a multicultural demographic population. Clearly, that poor matching rate is not sufficient to support data linkage for patient care, care coordination and research, as well as many other use cases, and would fare poorly in the very diverse New York City area. Therefore, New York City Clinical Delivery Research Network (NYC-CDRN) and Healthix chose to proceed with the IBM matching already proven and in use at Healthix. Not only did this approach save money, but it also saves time because the algorithm implementation and tuning had already been done.
What were the lessons learned thus far?
One of the discoveries was about the depth of data that established participants have been sending to Healthix for their core business activities. In reviewing data throughout the lifecycle of the CDRN activities it was discovered that some Healthix participants had only back loaded two years of data when they joined Healthix; whereas, others had sent entire patient demographic data loads. Having only two years of history would compromise some of the research objectives; thus, these organizations have now provided an incremental load of patient data that represents the comprehensive patient demographic history.
Has participation in the NYC-CDRN impacted Healthix in a positive or negative way?
Tom Check was pleased to share that Healthix participation in the NYC-CDRN has created great interest in the core business of Healthix, data exchange for care coordination, and additional participants have joined Healthix.
What happens next in the NYC-CDRN activities?
The NYC-CDRN participants are completing the next iteration of matching patients to clinical records and anonymizing them per the agreed upon protocols. In addition, the New York Genomics Center is working with the actual healthcare sites to receive and analyze the associated clinical records. This iteration is expected to be completed in early summer 2015. Additionally, the Patient-Centered Outcomes Research Institute (PCORI) is reviewing applications for phase two of the CDRN grants—ten were issued—and NYC-CDRN has submitted for additional funding and activities.